Distribution of firm size firms

---

Import libraries

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import plotly.io as pio
import plotly.express as px
import ipywidgets as widgets

pd.options.display.max_colwidth = 200
pd.set_option('display.max_columns', None)

---

Load Compustat data

df= pd.read_csv("data/compustat_from_1990.csv")
df['sic1']= (df['sich']/1000).astype(int).astype(str)
df=df[['conm', 'gvkey', 'tic', 'fyear', 'at', 'ni', 'sale', 'mv', 'sic1', 'sich']]
df['lag_at']= df.groupby('gvkey')['at'].shift(1)
df['ROA']= df['ni'] / df['lag_at']
print(df.shape)
df.head(3)

(211983, 12)

	conm	gvkey	tic	fyear	at	ni	sale	mv	sic1	sich	lag_at	ROA
0	AAR CORP	1004	AIR	1990	379.958	14.801	466.542	224.460360	5	5080.0	NaN	NaN
1	AAR CORP	1004	AIR	1991	395.351	10.020	422.657	204.699595	5	5080.0	379.958	0.026371
2	AAR CORP	1004	AIR	1992	365.151	0.283	382.780	214.663477	5	5080.0	395.351	0.000716

---

Todo

Define sich based on sic1 code

https://en.wikipedia.org/wiki/Standard_Industrial_Classification

---

df.dropna(inplace=True)
df= df[df['ROA'].abs()<0.5]
df= df[df['sale']>0]
df= df[df['at']>0]
df= df[df['mv']>0]
df.shape

(143046, 12)

---

Keep firms without missing obs and SIC change

df_N= df.groupby('gvkey').agg({'gvkey':'count', 'sic1':'nunique'}).rename(columns={'gvkey':'N', 'sic1':'ind_count'}).reset_index()
df_N.head()

	gvkey	N	ind_count
0	1004	30	1
1	1009	4	1
2	1011	4	1
3	1013	19	1
4	1017	4	1

df_to_filter= df_N[(df_N.ind_count==1) & (df_N.N==df_N.N.max())]
balance_gvkey_list= list(df_to_filter.gvkey)
len(balance_gvkey_list)

585

df= df[df.gvkey.isin(balance_gvkey_list)]
df.shape

(17550, 12)

df.fyear.value_counts().sort_index().plot(kind='bar')

<AxesSubplot:>

Balanced panel is confirmed.

---

Keep the largest 100 unique firms

df_2020= df[df.fyear==2020]
gvkey_100_list= list(df_2020.sort_values('mv', ascending= False).head(100)['gvkey'])
df=df[df.gvkey.isin(gvkey_100_list)]
df.shape

(3000, 12)

---

Visualization

xcol='sale'
ycol='ROA'

# Define size variable to be used as a bubble size
bubble_size_base= 'mv'
df['bubble_size']= df[bubble_size_base].clip(df[bubble_size_base].quantile(0.3),)

fig=px.scatter(df, x=xcol, y=ycol,
              animation_frame='fyear',
              animation_group='conm',
              height= 600, width= 800,
              range_x=[min(df[xcol]), max(df[xcol])],
              range_y=[min(df[ycol]), max(df[ycol])],
               size='bubble_size', 
               size_max= 50,
               log_x= True,
               color='sic1',
              hover_name= 'conm'
              )

fig.update_layout(
    title={
        'text': f"Dynamic Scatter Plots: {xcol} & {ycol} with {bubble_size_base} as a bubble size",
        'x':0.5,
        'xanchor': 'center',
        'yanchor': 'top'})

fig.show()

xcol='mv'
ycol='at'

# Define size variable to be used as a bubble size
bubble_size_base= 'mv'
df['bubble_size']= df[bubble_size_base].clip(df[bubble_size_base].quantile(0.3),)

fig=px.scatter(df, x=xcol, y=ycol,
              animation_frame='fyear',
              animation_group='conm',
              height= 600, width= 800,
              range_x=[min(df[xcol]), max(df[xcol])],
              range_y=[min(df[ycol]), max(df[ycol])],
               size='bubble_size', 
               size_max= 50,
               log_x= True, # Log transformation for x
               log_y= True, # Log transformation for y
               color='sic1',
              hover_name= 'conm'
              )

fig.update_layout(
    title={
        'text': f"Dynamic Scatter Plots: {xcol} & {ycol} with {bubble_size_base} as a bubble size",
        'x':0.5,
        'xanchor': 'center',
        'yanchor': 'top'})

fig.show()

fig.write_html('dynamic-scatter-mv-at.html', auto_play= False)

---

! jupyter nbconvert compustat-bubble-plot-animation.ipynb --to html

[NbConvertApp] Converting notebook compustat-bubble-plot-animation.ipynb to html
[NbConvertApp] Writing 1288453 bytes to compustat-bubble-plot-animation.html

xcol='mv'
ycol='at'

# Define size variable to be used as a bubble size
bubble_size_base= 'mv'
df['bubble_size']= df[bubble_size_base].clip(df[bubble_size_base].quantile(0.3),)

fig=px.scatter(df, x=xcol, y=ycol,
              animation_frame='fyear',
              animation_group='conm',
              height= 600, width= 800,
              range_x=[min(df[xcol]), max(df[xcol])],
              range_y=[min(df[ycol]), max(df[ycol])],
               size='bubble_size', 
               size_max= 50,
               log_x= True, # Log transformation for x
               log_y= True, # Log transformation for y
               color='sic1',
              hover_name= 'conm'
              )

fig.update_layout(
    title={
        'text': f"Dynamic Scatter Plots: {xcol} & {ycol} with {bubble_size_base} as a bubble size",
        'x':0.5,
        'xanchor': 'center',
        'yanchor': 'top'})

fig.show()

app = JupyterDash(__name__)

app.layout= html.Div([
    html.H4('TITLE'),
    dcc.Dropdown(['asset', 'sale']),
    dcc.Graph(id='scatter-plot'),
    html.P("Select X:")    
])

@app.collaback(
    Output("scatter-plot", "figure")
)

def scatter_plot(slider_range):
    
    xcol='mv'
    ycol='at'

    # Define size variable to be used as a bubble size
    bubble_size_base= 'mv'
    df['bubble_size']= df[bubble_size_base].clip(df[bubble_size_base].quantile(0.3),)        

    
    fig=px.scatter(df, x=xcol, y=ycol,
                  animation_frame='fyear',
                  animation_group='conm',
                  height= 600, width= 800,
                  range_x=[min(df[xcol]), max(df[xcol])],
                  range_y=[min(df[ycol]), max(df[ycol])],
                   size='bubble_size', 
                   size_max= 50,
                   log_x= True, # Log transformation for x
                   log_y= True, # Log transformation for y
                   color='sic1',
                  hover_name= 'conm'
                  )
    return fig


app.run_server(mode='inline')

from dash import Dash, dcc, html, Input, Output
import plotly.express as px
from jupyter_dash import JupyterDash

app = JupyterDash(__name__)
app.layout = html.Div([
    html.H4('Interactive scatter plot with Iris dataset'),
    dcc.Graph(id="scatter-plot"),
    html.P("Filter by petal width:"),
    dcc.RangeSlider(
        id='range-slider',
        min=0, max=2.5, step=0.1,
        marks={0: '0', 2.5: '2.5'},
        value=[0.5, 2]
    ),
])
@app.callback(
    Output("scatter-plot", "figure"), 
    Input("range-slider", "value"))
def update_bar_chart(slider_range):
    df = px.data.iris() # replace with your own data source
    low, high = slider_range
    mask = (df['petal_width'] > low) & (df['petal_width'] < high)
    fig = px.scatter(
        df[mask], x="sepal_width", y="sepal_length", 
        color="species", size='petal_length', 
        hover_data=['petal_width'])
    return fig


app.run_server(mode='inline')